【作者】 杜健辉；

【导师】 王国荣；

【作者基本信息】 华南理工大学 ， 材料加工工程， 2011， 博士

【摘要】 随着海洋资源的开发,海洋工程的建造规模和施工深度在不断加大。水下焊接作为海洋工程建造和维修中必不可少的关键技术,一直是各国进行先进制造技术研发的重点。因潜水焊工受到饱和潜水深度的限制,迫切需要应用机器人来取代潜水焊工,实现水下焊接的自动化和智能化。基于旋转电弧传感的焊缝自动跟踪系统直接使用焊接过程中的焊接电流信号进行焊缝跟踪。旋转电弧传感器引起焊丝在V形坡口上偏心旋转,由于V形坡口上高度的不同引起焊接弧长发生有规律的变化,焊接电流发生相应的变化。通过分析焊接电流信号的变化规律来获得焊缝偏差信息,从而实现焊缝自动跟踪。旋转电弧传感器的结构简单,稳定性好,灵敏度高,空间可达性好,且成本低,因而它是目前最有效的焊缝跟踪方法之一。旋转电弧传感器应用于水下焊缝自动跟踪免除了视觉系统带来的不便、麻烦,显示出其优越性,但与陆上相比,水下电弧更不稳定,信号获取、处理和偏差识别均带来更多的困难,所以有必要对旋转电弧传感器的水下自动跟踪进行研究。本文首先参考原有旋转电弧传感器的结构,在此基础上进行结构小型化和水下适应性改进,并设计与微型排水罩进行连接。以改进后的电弧传感器为基础建立一套基于旋转电弧传感的水下焊缝跟踪系统平台,为进行水下焊缝跟踪试验奠定研究基础。该跟踪系统包括旋转电弧传感器、电机调速控制系统、霍尔电流传感器、数据采集卡、工控机和机器人控制系统。利用Visual C++ 2005开发平台开发了水下焊缝跟踪软件系统,它包括焊接模块,信号采集模块,直流电机调速控制模块,信号处理模块,焊缝偏差识别模块和机器人控制模块。由于旋转电弧传感器水下焊接过程受多因素相互作用影响,需进行大量试验才能获得最佳工艺参数。通过正交试验的方法,以焊缝成形和电流信号特征的好坏作为评价指标,用较少实验次数确定了优焊接参数,为后续水下焊缝跟踪提供实验依据。同时测量焊缝成形的工艺参数,通过建立水下焊缝几何尺寸的回归方程,对影响熔宽和熔深的各主要因素进行敏感性分析。运用智能学习方法,建立水下焊缝几何尺寸的预测模型。由于焊接过程中存在强烈的弧光、飞溅、烟尘等干扰,采集的焊接电流信号常受到高频噪声、熄弧和弧光干扰。通过对比各种滤波方法,采取小波滤波和中值滤波组合滤波的方法进行滤波,试验结果表明组合滤波算法能有效的去除信号干扰,保留信号波形特征,得到更直观的水下焊接电流信号。提出了主成分分析和关联向量机结合的水下焊缝偏差识别算法。对基于旋转电弧传感的水下焊接电流信号构建的焊缝偏差样本数据集进行主成分分析,有效降低数据维数并消除数据间自相关性;通过关联向量机对降维后数据进行训练并进行识别实验。实验结果表明,结合主成分分析和关联向量机的算法加快了关联向量机的运算速度,精度与普通的关联向量机算法相差不大,比区间积分法、神经网络法和支持向量机法更高;在运行速度上,虽比区间积分法慢,但比神经网络法、支持向量机和普通的关联向量机更快。最后利用基于旋转电弧传感的焊缝跟踪系统进行陆上、湿法和微型排水罩的跟踪实验,工控机与机器人控制系统的实时串口通信,实现斜线和S形曲线焊缝的陆上、水下湿法和基于微型排水罩焊缝跟踪试验。跟踪实验结果表明本文研究的机器人水下焊缝自动跟踪系统是成功的,微型排水罩的局部干法成形质量和跟踪效果优于水下湿法焊接,更适合实际工程应用。更多还原

【Abstract】 The scale of ocean project becomes larger with the exploiting of marine resources. As a key technology of ocean project construction and maintenance,underwater welding is one of the main points in advanced technology of manufacturing . Due to the restrict of underwater welder in deep water, underwater robot, instead of underwater welder, is in urgent need to excute automatic underwater welding.The underwater robot seam tracking system based on rotating arc sensor track seam with welding current signal. When rotating arc sensor is rotating above the V groove, the offset between the center of the welding torch and the middle of seam can be identified for arc length and welding current change in accordance. Seam tracking system based on rotating arc sensor is one of the most effective methods to seam tracking for simple structure, excellent accessibility and lower cost.Compring to the visual sensor, underwater rotating arc sensor is more suitable for underwater welding for its advantages. It is difficult to acquire and process the signal for the arc in underwater welding is more unstable than in land, it is need to study underwater seam track system based on rotating arc sensor.The minitype rotating arc sensor is designed for improving accessibility and reducing vibration,and the structure of mini drain cap. Hence, underwater seam tracking system based on the rotating arc sensor is built for experiment. The system contains rotating arc sensor, speed control system, current sensor, data acquired card, computer and robot control system. The underwater seam tracking software system is developed by Visual C++ 2005, includes signal capture, speed control module, signal processing, offset identification, and serial port communication.The underwater welding based on rotating arc sensor is affected by many factors, it is necessary to do a great deal of experiments to get the optimized technological parameters. The orthogonal experiment with four major factors is designed to obtain the optimized technological parameters on welding forming and characteristic of welding signal, which provide experiment basis for the seam tracking. Then the regression equation which is built form the geometry of underwater welding is used to periform sensitivity analysis.RVM (Relevance vector machine) is used to build a model to predict the geometry of underwater rotating welding.Due to the disturbance of the arc light, splash, soot and so forth, the acquired signal often is disturbed by the break circuit and noise. Wavelet filter and median filter is used to process the welding signal by comparing some denoise algorithm. The result shows that the filter method can filter the disturb, keep the characteristic of welding signal, and obtain more direct signal.The underwater seam offset identification method that combine PCA(Principal component analysis) and RVM is presented.The principal component analysis is usded to process the data set which acquired in the underwater welding experiment, which can reduce the dimension of data set. Then the processed data is trained in the RVM.The result shows that, the precision of PCA_RVM is as good as RVM, and better than interval integral method, neural network and support vector machine; The runtime of PCA_RVM is more than interval integral method and less than neural network,support vector machine and RVM,PCA_RVM is more suitable for seam tracking system.Finally a set of land、wet and local dry with mini drain cap welding experiments are performed on the steel plate with v-groove. The seam tracking system is used to track oblique line and s-curve in land、wet and local dry with mini drain cap. The results of the tracking experiments show that seam tracking system is successful.local dry with mini drain cap underwater welding have better welding quality and tracking result than wet underwater welding ,and more suitable for engineering applications.更多还原